Adrian Hauswirth

Adrian Hauswirth

Auto-generated profile, based on publicly available information

Adrian Hauswirth

Department of Information Technology and Electrical Engineering ETH Zürich Zurich Switzer- land

Research Focus & Metrics

Academic Career Length

12Years (2015-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 April 2020 in IEEE Transactions on Automatic Control
Verified
Journal Article
Closed access
AHAdrian Hauswirth
Adrian Hauswirth
SBSaverio Bolognani
Saverio Bolognani
GHGabriela Hug
Gabriela Hug
FDFlorian Dorfler
Florian Dorfler
Autonomous optimization refers to the design of feedback controllers that steer a physical system to a steady state that solves a predefined, possibly constrained, optimization problem. As such, no exogenous control inputs such as setpoints or trajectories are required. Instead, these controllers are modeled after optimization algorithms that take the form of dynamical systems. The interconnection of this type of optimization dynamics with a physical system is however not guaranteed to be stable unless both dynamics act on sufficiently different timescales. In this paper, we quantify the required timescale separation and give prescriptions that can be directly used in the design of this type of feedback controllers. Using ideas from singular perturbation analysis, we derive stability bounds for different feedback laws that are based on common continuous-time optimization schemes. In particular, we consider gradient descent and its variations, including projected gradient, and Newton gradient. We further give stability bounds for momentum methods and saddle-point flows. Finally, we discuss how optimization algorithms like subgradient and accelerated gradient descent, while well-behaved in offline settings, are unsuitable for autonomous optimization due to their general lack of robustness.

1 June 2017
Verified
Conference Paper
Closed access
AHAdrian Hauswirth
Adrian Hauswirth
AZAlessandro Zanardi
Alessandro Zanardi
SBSaverio Bolognani
Saverio Bolognani
FDFlorian Dorfler
Florian Dorfler
GHGabriela Hug
Gabriela Hug
The focus of this paper is the online load flow optimization of power systems in closed loop. In contrast to the conventional approach where an AC OPF solution is computed before being applied to the system, our objective is to design an adaptive feedback controller that steers the system in real time to the optimal operating point without explicitly solving an AC OPF problem. Our approach can be used for example to simultaneously regulate voltages, mitigate line congestion, and optimize operating costs under time-varying conditions. In contrast to related work which is mostly focused on distribution grids, we introduce a modeling approach in terms of manifold optimization that is applicable in general scenarios. For this, we treat the power flow equations as implicit constraints that are naturally enforced and hence give rise to the power flow manifold (PFM). Based on our theoretical results for this type of optimization problems, we propose a discrete-time projected gradient descent scheme on the PFM. In this work, we confirm through a detailed simulation study that the algorithm performs well in a more realistic power system setup and reliably tracks the time-varying optimum of the underlying AC OPF problem.

1 September 2016
Verified
Conference Paper
Closed access
AHAdrian Hauswirth
Adrian Hauswirth
SBSaverio Bolognani
Saverio Bolognani
GHGabriela Hug
Gabriela Hug
FDFlorian Dorfler
Florian Dorfler
Motivated by online optimization problems arising in nonlinear power system applications, this article concerns optimization over closed subsets of Riemannian manifolds. Compared to conventional optimization over manifolds, we explicitly consider inequality constraints that result in a feasible set that is itself not a smooth manifold. We propose a continuous-time projected gradient descent algorithm over the feasible set and show its well-behaved convergent behavior. Under mild assumptions on the non-degeneracy of equilibria we show that points are local minimizers if and only if they are asymptotically stable. The proposed algorithm can be implemented as a real-time feedback control law on a physical system. This approach is particularly appropriate for online load flow optimization problem in power systems, in which the state of the grid is naturally constrained to the manifold that represents the solution space to the nonlinear AC power flow equations. We specialize our approach for the case of power distribution systems that need to respect operational constraints while being economically efficient, and we illustrate the resulting closed-loop behavior in simulations.

1 December 2020 in Electric Power Systems Research
Verified
Journal Article
Closed access
LOLukas Ortmann
Lukas Ortmann
AHAdrian Hauswirth
Adrian Hauswirth
ICIvo Caduff
Ivo Caduff
FDFlorian Dörfler
Florian Dörfler
SBSaverio Bolognani
Saverio Bolognani

1 January 2021 in SIAM Journal on Control and Optimization
Verified
Journal Article
Closed access
AHAdrian Hauswirth
Adrian Hauswirth
SBSaverio Bolognani
Saverio Bolognani
FDFlorian Dörfler
Florian Dörfler
Continuous-time projected dynamical systems are an elementary class of discontinuous dynamical systems with trajectories that remain in a feasible domain by means of projecting outward-pointing vec...

21 June 2018 in IEEE Control Systems Letters
Verified
Journal Article
Closed access
AHAdrian Hauswirth
Adrian Hauswirth
SBSaverio Bolognani
Saverio Bolognani
GHGabriela Hug
Gabriela Hug
FDFlorian Dorfler
Florian Dorfler
Solutions to nonlinear, nonconvex optimization problems can fail to satisfy the KKT optimality conditions even when they are optimal. This is due to the fact that unless constraint qualifications (CQ) are satisfied, Lagrange multipliers may fail to exist. Even if the KKT conditions are applicable, the multipliers may not be unique. These possibilities also affect AC optimal power flow (OPF) problems which are routinely solved in power systems planning, scheduling and operations. The complex structure – in particular the presence of the nonlinear power flow equations which naturally exhibit a structural degeneracy – make any attempt to establish CQs for the entire class of problems very challenging. In this paper, we resort to tools from differential topology to show that for AC OPF problems in various contexts the linear independence constraint qualification is satisfied almost certainly, thus effectively obviating the usual assumption on CQs. Consequently, for any local optimizer there generically exists a unique set of multipliers that satisfy the KKT conditions.